In the literature, the acid earth metals niobium and tantalum in particular are described as starting materials for the production of such capacitors. The capacitors are produced by sintering of the finely divided powders to pellets to produce a structure having a large surface area, anodic oxidation of the surface of those sintered bodies to produce a non-conducting insulating layer (dielectric), and application of the counter electrode in the form of a layer of manganese dioxide or of a conductive polymer. The particular suitability of acid earth metal powders is derived from the high relative permittivity of the pentoxides.
Hitherto, only tantalum powder has gained industrial importance for the production of capacitors. That is based on the one hand on the reproducible producibility of finely divided tantalum powder and, on the other hand, on the fact that the insulating oxide layer of tantalum pentoxide possesses particularly pronounced stability. That is possibly due to the fact that tantalum, unlike niobium, does not form a stable suboxide.
In the course of the rapid development of microelectronics, however, disadvantages of tantalum as to its availability and some of its features are increasingly gaining importance. Tantalum is one of the rare metals (54th position in the natural frequency of the elements in the earth's crust with 2.1 g/t) with few mineable deposits (only hard rock mining) and, moreover, it is found in only very small concentrations in its ores. For example, the tantalum ores typically mined today (e.g. in Australia) often contain less than 0.1% Ta2O5 (approx. 300 ppm Ta). Niobium, which is in the same group of the PSE above tantalum and is very similar thereto in terms of its behaviour, occurs from 10 to 12 times more frequently than tantalum and its deposits are more favourably mineable (33rd position in the natural frequency of the elements in the earth's crust with 24 g/t). The most important deposits in commercial terms are in Brazil (78% of world reserves), where the ore is mined in opencast pits with over 3% Nb2O5. Further deposits are to be found in Canada, Nigeria and Zaire. Accordingly, the raw material prices for niobium ore concentrates are markedly lower than for tantalum ore concentrates and, moreover, are not subject to such pronounced fluctuations.
Furthermore, there is a natural growth limit to the achievable specific capacitances for tantalum powder. In order to achieve higher capacitances C in the case of Ta powder, the specific surface area must become larger (C=∈o∈r*A/d), which at a particular powder particle geometry is accompanied by a reduction in the size of the particles. If the mean particle size, in the case of an anodically produced dielectric layer in the nanometer range, is likewise in the nanometer range, regions of the metal sintered body become “through-anodised”. That is to say there is no metallic conductivity between two particles, particularly in thin areas such as, for example, sinter necks. Parts of the anode thus become inactive.
Moreover, the sensitivity of tantalum powders to oxidation increases markedly as the size of the powder particles decreases and the specific surface area increases accordingly.
For those reasons, and owing to the markedly higher dielectric constants of niobium (∈r˜42) as compared with tantalum (∈r˜27), it has been the aim of many researchers to develop niobium capacitors. However, the use of niobium capacitors has hitherto been reserved for the field of low specific capacitances with a small specific surface area and relatively poor quality. One reason therefor is that pure niobium has two disadvantages in comparison with tantalum with regard to capacitor applications. On the one hand, the tendency of the anodically produced oxide film to field crystallisation is more pronounced than in the case of tantalum. The radial growth rate of crystalline surfaces is, in fact, 1000 times greater than in the case of tantalum under the same conditions of anodisation (N. F. Jackson, J. C. Hendy, Electrocomponent Science & Techn. 1974, 1, 27-37). This can, however, for the most part be suppressed by anodisation at a lower temperature (Y. Pozdeev: “Comparison of tantalum and niobium solid electrolytic capacitors” TIC 1997; films must be amorphous, crystalline areas in the film exhibit increased conductivity). The other disadvantage concerns the greater sensitivity of anodically produced Nb2O5 films to heat treatment.
One step in the production of solid electrolytic capacitors is the application of the semiconducting cathode material MnO2. That is effected by immersing the anode body in manganese nitrate solutions to produce a thin MnNO3 layer, which is subsequently decomposed thermally to MnO2. In that process, the Ta—Ta2O5 system is exposed to temperatures of from 250 to 450° C. for from 10 to 30 minutes. Such heat treatment may, however, lead to an increase in the frequency-, temperature- and BIAS-dependence of the capacitance. The cause thereof is considered to be that, at temperatures above 300° C., the tantalum substrate is able to draw oxygen atoms from the anodically produced tantalum oxide layer, which leads to an exponential gradient of areas in the oxide film that lack oxygen. Such lacking areas bring about a change in the conducting behaviour of the oxide film from a dielectric to an n-type semiconductor or, if the lacking areas are present in a sufficiently high concentration, to a conductor. That is shown diagrammatically in FIG. 1. The critical conductivity σ0 separates the insulating part of the oxide film from the conducting part. If the temperature is increased, the semiconducting layer in the oxide film widens and the effective insulating layer becomes thinner. That causes an increase in capacitance, independently of the temperature-dependence of the dielectric constant. In such a case, the application of an anodic BIAS voltage causes the electrons to move from the lacking areas into the tantalum metal. This results in the formation of an electric double layer, which is defined on the metal side by electrons at the interface and on the semiconductor side by the positive space charge in a boundary layer low in charge carriers (Schottky-Mott barrier). That effects an increase in the gradient of the conductivity gradient and an increase in the effective thickness of the dielectric, which, however, according to C=∈o∈r*A/d, is associated with a reduction in the capacitance.
While anodically produced oxide films on tantalum are dielectric and exhibit semiconducting regions only at elevated temperatures, anodically produced oxide films on niobium behave like n-type semiconductors even at room temperature (A. D. Modestov, A. D. Dadydov, J. Electroanalytical Chem. 1999, 460, pp. 214-225). And, they exhibit a Schottky-barrier at the Nb2O5/electrolyte interface (K. E. Heusler, M. Schulze, Electrochim. Acta 1975, 20, p. 237; F. Di Quarto, S. Piazza, C. Sunseri, J. Electroanalytical Chem. 1990, 35, p. 99). The reason therefor may be that niobium, in contrast to tantalum, forms various stable suboxides. For example, it is known from the literature that, in the case of oxide films on niobium, only the outer layer consists of Nb2O5-x (M. Grundner, J. Halbritter, J. Appl. Phys. 1980, 51(1), pp. 397-405), which moreover, is not completely stoichiometric in composition and exhibits an oxygen deficiency x. Between the Nb2O5-x layer and the niobium metal substrate there is a layer of NbO, since that is the thermodynamically stable phase in contact with the oxygen-saturated niobium metal and not, as in the case of tantalum, the pentoxide (K. E. Heusler, P. Schlüter, Werkstoffe & Korrosion 1969, 20(3), pp. 195-199).
The oxygen content of the passive surface layer in the case of niobium is approximately from 3500 to 4500 ppm per m2 specific surface area. When Nb anodes are sintered, the oxygen of the passive surface layer diffuses into the inside of the metal and is uniformly distributed therein. In that process, the thickness of the NbO layer also increases proportionally to the surface area of the powder used, which can very readily be followed on sintered niobium anodes by means of X-ray diffraction. In an extreme case, with very high specific surface areas and accordingly very high oxygen contents in the powder, the result is that the anode body consists mainly of NbO after sintering and not of niobium metal. In contrast to tantalum, however, that oxygen increase does not manifest itself in a significant rise in the residual current of anodes made of such powders.
A further point is that the MnO2 cathode acting as the solid electrolyte acts as an oxygen donor and is able to compensate for the oxygen deficit in the Nb2O5-x layer. That is not a monotonous process, however, since lower, non-conducting manganese oxide phases (Mn2O3, Mn3O4, MnO) form in the vicinity of the MnO2/Nb2O5 interface and suppress the further diffusion of oxygen from the MnO2 cathode to the semiconducting Nb2O5-x layer. That then leads to an increase in the lacking areas x, an accelerated rise in the residual current and, finally, to the failure of the capacitor (Y. Pozdeev on CARTS-EUROPE '97: 11th European Passive Components Symposium). For that reason, niobium capacitors are said to have a markedly shorter life than tantalum capacitors.
That semiconducting behaviour of the anodically produced barrier layer on niobium has the result that, in order to measure correct capacitance values for niobium anodes, which are later achieved also in the finished capacitor, a positive BIAS voltage must be applied thereto. Otherwise a meaningful measurement is not possible and values are simulated that are much too high.
By comparative measurements of the capacitance of anodes of niobium metal or niobium(II) oxide and also niobium/tantalum alloys (90:10, 80:20, 70:30) and the capacitors produced therefrom, it has been found that the application of a BIAS voltage of ≧1.5 V at the anode is necessary in order to measure for the anodes correct capacitance values, which arc also found again later in the finished capacitor, and that capacitances of such anodes measured without an applied BIAS voltage are higher by a factor of from 3 to 4 than those measured with a BIAS voltage of at least 1.5 V, that is to say incorrect values are simulated. Accordingly, values are also obtained for the specific residual current that are lower by a factor of from 3 to 4 than the actual specific residual current when reference is made to capactances measured without BIAS.
A very important parameter for the suitability of a powder as capacitor material is its chemical purity, since both metallic and non-metallic impurities can lead to faults in or to reduced stability of the dielectric layer. The elements Na, K, Fe, Cr, Ni and C in particular are to be regarded as critical for the residual current of tantalum anodes. As a result of continuous improvements to Ta powders, such impurities in powders produced by sodium reduction of K2TaF7 are nowadays in the region of the detection limit.
By contrast, the corresponding process via K2NbF7 is not available for the production of highly pure niobium powders because, owing to the high aggressivity of the corresponding heptafluoroniobate salts, the retort material is partly dissolved and the niobium powders so obtained are contaminated with large amounts of Fe, Cr, Ni, etc. So-called EB powders, which are produced by embrittling with hydrogen a niobium ingot melted by means of an electron beam, grinding it and subsequently dehydrating it, are also not suitable for the production of high-capacitance Nb capacitors. If the above-described grinding is carried out in an attritor under, for example, alcohols, niobium flakes are obtained which, however, in most cases contain a very high degree of metallic impurities, such as Fe, Cr, Ni and C, which are trapped in the niobium powder during the grinding operation by mechanical alloying and cannot be washed out later with mineral acids.
However, a very high degree of purity is exhibited by the niobium powders obtained by published proposals of the Applicants according to DE 19831280 A1 or WO 00/67936 by the two-stage reduction of niobium pentoxide with hydrogen or gaseous magnesium. Such powders contain, for example, metallic impurities such as Fe, Cr, Ni, Al, Na, K in amounts <25 ppm.
In addition to chemical purity, which is of decisive importance for the electrical properties, a capacitor powder must also meet some requirements in respect of physical properties. For example, it must have a certain flowability, so that it can be processed using the capacitor manufacturers' fully automated anode presses. Furthermore, a certain green strength of the pressed anode bodies is necessary so that they do not immediately fall apart again, and a sufficiently high pore distribution is required in order to ensure complete impregnation with manganese nitrate.
The object of the present invention is to overcome the above-described disadvantages of the known capacitors based on niobium. In particular, it is the object of the present invention to improve the insulating behaviour and the thermal stability of the niobium pentoxide barrier layer of capacitors based on niobium in such a manner that longer lives associated with higher capacitances and lower residual currents can be achieved for such capacitors.
It has now been found that such capacitors based on niobium exhibit markedly improved properties of the anodically produced oxide film if at least the barrier layer is alloyed/doped with vanadium. In particular, it has been found with the aid of impedance spectroscopic measurements and evaluation of Schottky-Mott diagrams that the concentration of lacking areas in anodically produced oxide layers of such capacitor anodes is markedly reduced and similarly low as in corresponding Ta2O5 layers. Moreover, there are the first signs of long-term stability comparable with that of tantalum anodes, which cannot be achieved with conventional capacitors based on niobium.